diff options
| author | Frank Lei <frank.lei@arm.com> | 2018-02-01 14:47:14 +0800 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:47:40 +0000 |
| commit | 4406fd6cc4abded564d3791324e1f48bdfd34273 (patch) | |
| tree | 22fe402fe9ac7ca338df49e9eccd6eb1587ae875 /src/core/GLES_COMPUTE/cs_shaders/concatenate.cs | |
| parent | 898d399a0f62c15612a52df4bff5018e783214e4 (diff) | |
| download | ComputeLibrary-4406fd6cc4abded564d3791324e1f48bdfd34273.tar.gz | |
APPBROWSER-391: Fix GLES COMPUTE alignment issues
APPBROWSER-402: Performance optimization for squeezenet/xray model
Change-Id: If31b186b99a6d6087164019fe94d3ac9279e3204
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/119526
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Diffstat (limited to 'src/core/GLES_COMPUTE/cs_shaders/concatenate.cs')
| -rw-r--r-- | src/core/GLES_COMPUTE/cs_shaders/concatenate.cs | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/src/core/GLES_COMPUTE/cs_shaders/concatenate.cs b/src/core/GLES_COMPUTE/cs_shaders/concatenate.cs index 0c8b5bf0bb..69ac50b4d0 100644 --- a/src/core/GLES_COMPUTE/cs_shaders/concatenate.cs +++ b/src/core/GLES_COMPUTE/cs_shaders/concatenate.cs @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -53,8 +53,8 @@ void main(void) Tensor3DIterator src_iter = CONVERT_TO_TENSOR3D_ITERATOR(src_attrs, src_shift); Tensor3DIterator dst_iter = CONVERT_TO_TENSOR3D_ITERATOR(dst_attrs, dst_shift); - float tmp = LOAD(src_ptr, TENSOR3D_OFFSET(src_iter, -OFFSETS_X, -OFFSETS_Y, 0)); - STORE(dst_ptr, TENSOR_OFFSET_ADVANCE_IN_BYTES(dst_iter, OFFSETS_Z), tmp); + float tmp = LOAD(src_ptr, TENSOR3D_OFFSET(src_iter, -OFFSET_X, -OFFSET_Y, 0)); + STORE_CURRENT_ITEM(dst_ptr, dst_iter, tmp); } #elif defined(DATA_TYPE_FP16) @@ -66,7 +66,7 @@ void main(void) Tensor3DIterator src_iter = CONVERT_TO_TENSOR3D_ITERATOR(src_attrs, src_shift); Tensor3DIterator dst_iter = CONVERT_TO_TENSOR3D_ITERATOR(dst_attrs, dst_shift); - uvec2 tmp = LOAD(src_ptr, TENSOR3D_OFFSET(src_iter, -OFFSETS_X, -OFFSETS_Y, 0)); - STORE(dst_ptr, TENSOR_OFFSET_ADVANCE_IN_BYTES(dst_iter, OFFSETS_Z), tmp); + uvec2 tmp = LOAD(src_ptr, TENSOR3D_OFFSET(src_iter, -OFFSET_X, -OFFSET_Y, 0)); + STORE_CURRENT_ITEM(dst_ptr, dst_iter, tmp); } #endif /*DATA_TYPE_FP16*/ |